Closed bendichter closed 1 year ago
if a url can can be constructed with the source file url. the example you posted doesn't have that.
@satra, that testing example has the object ID hard coded. I'll implement a scheme where the URL contains the needed information.
So it would be something like this:
https://scratchrealm.github.io/neurosift/#/nwb/dandiarchive/c86cdfba-e1af-45a7-8dfd-d243adc20ced (DOESN'T WORK YET)
Would that be sufficient?
@magland - the way we have been doing it is through defining services like this: https://github.com/dandi/dandi-archive/blob/bbada9c42cf1ebffcd91a38b6a89d821f2391c4c/web/src/views/FileBrowserView/FileBrowser.vue#L307
which in turn creates urls like this in the file viewer.
that looks great. submitted a PR. will let kitware team take it over from here.
Thanks, I just transferred that repo to the flatironinstitute github org. So the above link will not work anymore, and you'll need to modify the PR.
It will now be:
(Although this won't work until tomorrow when I will get granted admin access.)
neurosift is a JS-based neurophysiology data visualization tool created by @magland. It allows for data to be streamed directly from DANDI (or any s3 bucket), and can automatically render visualizations from NWB data files. This is like NWB Widgets, except it is entirely client side visualizations, and does not require us to launch a jupyter kernel in the background.
Here is a WIP example of an automatic visualization of NWB.
We are still in the process of developing an ideal visualization and optimizing data storage to improve streaming performance. On the web UI side, would it be possible to add neurosift as an "external service" here?